Data Thinning of Species Occurrences in Environmental Space R package bean version 0.2.1 A suite of tools to mitigate sampling bias in species occurrence records by thinning data in the environmental space (E-space). This process can improve the accuracy and precision of species distribution models (SDM, also known as ecological niche models, ENM). The package offers a data-driven protocol to determine thinning parameters using kernel-density bandwidth selection. Two thinning methods are provided (stochastic and deterministic) to reduce over-sampled environmental conditions and down-weight outlier observations. The name 'bean' reflects the core principle of the method: each 'pod' (a grid cell in E-space) is allowed to contain only a limited number of 'beans' (occurrence points). See Silverman (1986, ISBN:978-0-412-24620-3) and Rousseeuw and Leroy (2003, ISBN:978-0-471-48855-2) for the underlying statistical methods. Software Depends: R (>= 4.0) Imports: MASS, stats, terra Suggests: covr, knitr, rmarkdown, testthat (>= 3.0.0), ggplot2, rgl Paanwaris Paansri <paanwaris@vt.edu> Comprehensive R Archive Network (CRAN) Paanwaris Paansri [cre, aut] (ORCID: <https://orcid.org/0000-0001-9992-098X>), Luis E. Escobar [aut, ctb] (ORCID: <https://orcid.org/0000-0001-5735-2750>) MIT + file LICENSE (https://CRAN.R-project.org/package=bean/LICENSE) 2026-06-08 application/tgz https://CRAN.R-project.org/package=bean doi:10.32614/CRAN.package.bean